Re: [lmi] actuarial tables format (was Re: Terse list of valuable projects)

[Václav Slavík]

Hi,

On 22 Mar 2012, at 13:05, Václav Slavík wrote:
> I'll write updated format proposal in another mail later.

What about using this for the three kinds of tables? I think the higher-level 
wrappers (<smoker>, <gender>, ...) are OK, so I leave them unchanged and omit 
here.

For 1D tables, I'd prefer to use explicit rows too -- both for consistency and 
because of Vadim's points about ease of editing:

    <aggregation>
        <value age="10">0.00106</value>
        <value age="11">0.00113</value>
        <value age="12">0.00148</value>
    </aggregation>

For select tables:

    <select period="3">
        <row age="10">
            <value>0.00106</value>
            <value>0.00140</value>
            <value>0.00165</value>
        </row>
        <row age="11">
            <value>0.00113</value>
            <value>0.00148</value>
            <value>0.00175</value>
        </row>
    </select>

The 'period' attribute is not strictly needed, but I suspect it will help with 
validation. If it won't, I'd remove it.

Finally, select-and-ultimate:

    <select-and-ultimate>
        <select period="3">
            <row age="10">
                <value>0.00106</value>
                <value>0.00140</value>
                <value>0.00165</value>
            </row>
            <row age="11">
                <value>0.00113</value>
                <value>0.00148</value>
                <value>0.00175</value>
            </row>
        </select>
        <ultimate>
            <value age="10">0.00106</value>
            <value age="11">0.00113</value>
            <value age="12">0.00148</value>
            <value age="13">0.00148</value>
        </ultimate>
    </select-and-ultimate>

The format of the <ultimate> element is -- and should be -- identical to 
<aggregate>'s content.

If this sounds roughly good, I'll start writing explicit schema for it and 
conversion code to dump used tables in this format. Both of these may trigger 
some changes to the format, so it's the next thing to do. 

BTW, speaking of which, would a generally usable tool to convert SOA binary 
tables into our format be useful? I assume we'll need at least a one-time-use 
tool to dump the tables used by LMI, but a hacked-together program would 
suffice for that, while anything that would be used more than once should be of 
higher quality.

Thanks,
Vaclav